def test_kmers(self):
"""
Iterate over all the possible 2-mers of a string
"""
s = "ATTGCTCA"
possible_2mers = list(kmers(s, 2))
self.assertEqual(possible_2mers,
['AT', 'TT', 'TG', 'GC', 'CT', 'TC', 'CA'])
def profile_random_kmer(profile, sequence):
"""
Generate a profile-randomly chosen k-mer in a sequence.
The size of the kmer will be deducted from profile.
:param profile: the profile
:param sequence: the sequence to search in
:return:
"""
k = len(profile)
distribution = tuple(
probability_from_profile(kmer, profile) for kmer in kmers(sequence, k))
index = biased_random(distribution)
return sequence[index:index + k]
def most_probable_kmer_from_profile(sequence, k, profile_matrix):
"""
Find the most profile-probable k-mer in a sequence
:param sequence: the sequence
:param k: the size of the k-mer
:param profile_matrix: the profile matrix
:return: the most probable k-mer in sequence according to the given profile matrix
"""
most_probable = (-1, None)
for kmer in kmers(sequence, k):
probability = probability_from_profile(kmer, profile_matrix)
if probability > most_probable[0]:
most_probable = (probability, kmer)
return most_probable[1]
def motifs_enumeration(sequences, k, d):
"""
Check if a motif of length k appears in each sequence in strings with at most d mismatches
:param sequences: the array of sequences
:param k: the length of the motif
:param d: the maximum number of mismatches
:return: the (k, d)-motifs in string as a set
"""
motifs = set()
for kmer in kmers(sequences, k):
neighborhood = neighbors(kmer, d)
for neighbor in neighborhood:
neighborhood2 = neighbors(neighbor, d)
if all(
any(neighbor2 in seq for neighbor2 in neighborhood2)
for seq in sequences):
motifs.add(neighbor)
return motifs
def greedy_motifs_search(sequences, k, cromwell=True):
"""
Tries to find a collection of motifs in a collection of sequences of DNA
:param sequences: the collection of sequences
:param k: the size of the motifs to search for
:param cromwell: should we use Cromwell's rule when generating the profile matrix?
:return: a collection of the most probable motifs (one motif for each sequence)
"""
best_motifs = None
for motif1 in kmers(sequences[0], k):
motifs = [motif1]
for sequence in sequences[1:]:
profile_matrix = profile(motifs, cromwell)
motifs.append(
most_probable_kmer_from_profile(sequence, k, profile_matrix))
if not best_motifs or motifs_entropy(motifs) < motifs_entropy(
best_motifs):
best_motifs = motifs
return best_motifs
def hamming_distance_two_strings(s1, s2):
"""
Compute the hamming distance between two strings.
If the second string if bigger than the first one, this will return the minimum hamming distance it finds
between the first k-mer, and all the possible k-mers in the second string.
Efficiency: O(nk) with k being the size of the first string, and n the size of the second
:param s1: the first string
:param s2: the second string
:return:
"""
k = len(s1)
min_distance = float("inf")
# We compute the hamming distance between seq1 and s for s being all the possibles strings the same size as seq1
# in seq2.
for s in kmers(s2, k):
distance = hamming_distance_same_size(s1, s)
if distance < min_distance: # If the current newly found distance is lower than the minimum we have now
min_distance = distance
return min_distance